Application of extended time-frequency domain average in ultrasonic detecting

Ultrasonic signal detection is essential for the ultrasonic-based applications such as ultrasonic flow measurements and nondestructive testing. The paper proposes three extended time-frequency domain average (ETFDA) techniques, which are based on the smoothed pseudo-Wigner-Ville distribution, continuous wavelet transform and Hilbert-Huang transform. These techniques combine beneficial time-frequency localization characteristics of the time-frequency analysis and abilities of the time domain averaging (TDA) to suppress noise interference. They are thus well adapted for detection of the ultrasonic signals even when they are strongly smeared by the noise or distorted in the medium. A number of tests conducted on simulated and actual ultrasonic signals have demonstrated that ETFDA provides a solid performance.